1. Introduction

Continuing trends of increasing speed and coverage in the mobile communications services marketplace indicate further deployment of higher-capacity communications technologies well into the foreseeable future. A major hurdle in the development of these networks will be the realization and implementation of reliable low-cost network interface nodes between high-bandwidth wireless networks and the fixed core communications infrastructure. As an example of ongoing development trends, envelope detection of a 10 Gbit/s amplitude shift key (ASK) modulated 110 GHz carrier wave employing microwave components was recently reported in [1]. It is anticipated that backhaul infrastructure for cellular and broadband wireless networks will converge with that of fixed broadband networks in the near future.

All-optical envelope detection allows easy integration of microwave wireless communications schemes with fixed optical access links such as fiber-to-the-customer-premises (FTTCP) infrastructure. Although most existing wireless communication systems employ heterodyning methods for radio-frequency (RF) down-conversion and detection, envelope detection of microwave signals has recently received attention due to its simplicity. Furthermore, simple, robust base station design and construction is an important requirement for the development of high bandwidth, high carrier-frequency wireless communications networks as these are characterized by smaller coverage cell size and therefore require that more base transceiver stations (BTS) be deployed over a given area. The authors have also reported successful envelope detection of a 3.25 Gbit/s signal stream, ASK modulated onto a 38 GHz carrier, by employing all-optical half-wave rectification followed by slow-bandwidth photodetection [2], to yield the downconverted baseband signal in the electrical domain.

In this paper, we report on an all-optical envelope detection approach employing external injection of a half-wave-rectified optical pass-band signal into a multi-quantum well (MQW) DFB laser which delivers the corresponding down-converted baseband signal in the optical domain. We propose to deploy this device at an intermediate point of an optical access network, to perform optical signal processing required to convert RF-over-Fiber (RFoF) signaling to baseband, and thus avoid the dispersion penalty incurred when RF carrier frequency signals are transmitted over the optical network [3]. The all-optical envelope detection simultaneously eliminates the requirement for more complex coherent detection systems at RF transceiver nodes. This all-optical scheme would therefore be integrable with a wireless network BTS connected to a FTTCP link, and presents the added benefit of lowering system complexity, representing potential cost savings for network implementation and maintenance. All-optical envelope detection was successfully performed with a 5.5 Gbit/s ASK system operating at 20 GHz carrier frequency.

2. Principle of operation

We consider the case of a base station employing optical envelope detection as depicted in Fig. 1. The ASK-modulated microwave signal received by the antenna drives an electro-absorption modulator (EAM), biased such that all-optical half-wave rectification is achieved [2]. This signal was directly injected into the active region of the DFB laser device and the resulting DFB output was evaluated to assess its suitability for data transport.

 

Fig. 1. All-optical envelope detection of wireless signals. EAM: electro-absorption modulator. DFB: distributed feed-back laser. V_bias: bias voltage

Download Full Size | PDF

It is well known [4, 5] that introducing photons into an active laser cavity results in decreased output optical power at the lasing wavelength due to gain suppression. This effect may be described as a cross-gain modulation (XGM) and has typically been regarded as a pernicious feature of laser operation: it is traditionally mitigated with optical isolation at the laser output. We instead seek to exploit this optical XGM effect in the development of a novel all-optical envelope detection scheme. Optical injection of a distributed Bragg reflector laser had previously been reported to perform high-speed optical packet detection with payload duration between 8 ns and 300 ns [5]. We instead wished to directly integrate the all-optical detection phenomenon directly into the data stream and evaluated the potential for replacing conventional optical signal conditioning circuitry with an appropriately biased DFB device.

3. Experimental setup

The experimental setup used for characterizing the gain suppression effect of an optical injected DFB laser for all-optical envelope detection is presented in Fig. 2. The core of this system is the external optical injected device, in this case implemented with a multiple-quantum well (MQW) InGaAs DFB device. This DFB laser is a commercially available, fiber pigtailed coax packaged device, with no optical isolation at the output; it is operated without temperature or wavelength stabilization circuitry. The central emitting wavelength λ_C was observed to lie around 1551 nm. A sample of the measured optical spectrum is presented in Fig. 3. The DFB laser used in these experiments was biased above threshold and optical signals injected into the active medium. The XGM effect caused laser output to be suppressed or allowed depending on the injected optical signal envelope. Since the laser response is slow compared to the radio-frequency carrier but faster than the source data rate, the resulting DFB output waveform therefore followed the inverted envelope of the optical source signal.

 

Fig. 2. Layout of the experimental setup. OBPF means optical band pass filter, MZM means Mach-Zehnder Modulator. Signal monitoring point (MP) locations are also shown.

Download Full Size | PDF

 

Fig. 3. Measured optical emission spectrum of the DFB laser used for external optical injection. The vertical axes have been normalized for convenient comparison. The peak wavelength λ_C was found to be close to 1551 nm; the first right side mode peak (λ_HR1) at 1553 nm. Higher side modes were observed at a periodic spectral separation of approximately 1.2 nm. The figure on the right indicates the spectrum observed at MP2 when the 2.5 GHz ASK signal is injected into the DFB at λHR1 and filtered by the OBPF.

Download Full Size | PDF

For experimental purposes, the optical transmitter was implemented by a pair of cascaded Mach-Zehnder interferometer modulator (MZM) units. The first MZM imposed the 20 GHz radio-frequency (RF) carrier. The second MZM imposed ASK modulation onto this optical carrier: the driving electrical signal was obtained from the pulse pattern generator. For RFoF input, the optical signal generated by the cascaded MZM is the same type obtained when an EAM device is used to perform half-wave rectification of an input RF passband signal [2]. This facilitated the evaluation of system response to baseband (by disabling the 20 GHz carrier wave) and passband RFoF (by enabling this carrier) signaling as required to emulate input from various types of terminal node equipment. A variable optical attenuator was used to set the value of the average optical power level injected into the DFB laser via an optical circulator. An optical band-pass filter (OBPF) was inserted after the EDFA to reduce system noise; the second OBPF allowed selection of the DFB emitting wavelength. Photodetection was implemented using a 40 GHz photodiode. The resulting electrical signal was filtered using a 1.8 GHz Bessel lowpass filter. Post-detector bit-error-rate performance was assessed using a bit-error-rate tester. Optical power splitters allowed monitoring and measurement of the signal at various Monitoring Points (shown). Polarization control was used to optimize system performance. Further work is required to optimize material selection and laser device design to reduce the polarization sensitivity of the cross-gain modulation effect at the core of this system and hence increase its attractiveness for practical implementation.

4. Results

First, we report on the experimental assessment of the DFB laser response to external injected optical signals with various wavelengths and power levels when operating at different driving current levels (static operation characterization). In this way, a suitable operating point was determined for the operation of the DFB as envelope detector. We then assessed the dynamic behavior of the system (data signal response) using PRBS of varying lengths and report on the bit-error rate measurements for the following cases:

We also evaluated the signal BER sensitivity under the following conditions:

Analysis was also done to determine the wavelength-dependency of the XGM effect exploited. Our observations are presented below.

4.1 Characterization of static operation

By disabling the two MZI modulators we consider the static case for different values of the tunable laser source emitting wavelength and input optical power level into the DFB. An example of the measurements results for the DFB output power and lase mode wavelength (λ_C) are presented in Fig. 4.

Four settings of the driving current were considered; all bias points selected were above the device threshold current value of 15 mA. Observations were made of the effect of operating point proximity to threshold and how this affected DFB output power and spectrum. Optical filtering was applied to remove signals at the operating wavelength of the tunable laser (TL). The DFB output power was measured when λ_TL was varied over a wide range of wavelengths around λ_C.

 

Fig. 4. Relationship between input optical power and DFB laser output power (black) and lasing wavelength (circles, blue) at various operating points (color online).

Download Full Size | PDF

It was found that the device input/output relationship exhibited a sharper response to variations to external optical input power when the TL was aligned with one the spectral peaks observed in Fig. 3, and this response was greatest when the TL operating wavelength was greater than λ_C. At TL operation with wavelength below λ_C, more injected power was required to produce a similar extinction of the DFB output. From the static measurement results seen in Fig. 4, the following trends were observed:

4.2 Data signal response

Consider an active transmission system implementing ASK modulation of a 20 GHz carrier, presenting optically halfwave rectified signals previously described [2]. We applied this 2.5 Gbit/s ASK-modulated output waveform to the DFB as shown in Fig. 2 and observed the signal characteristics of the test system. The power meter/variable attenuator at MP1 (see Fig. 2) was used to maintain a constant average optical power level injected into the DFB laser. Eye diagrams are presented as Figs. 5–6.

Figure 5 shows the signal propagation through the system with the carrier disabled: this corresponds to a baseband transmission scenario. The output from the optical transmitter (MP1) is presented in Fig. 5(b); the optical signal output from the DFB laser (MP2) is shown in Fig. 5(c); and the electrical waveform obtained after photodetection and low-pass filtering (MP3) is presented as Fig. 5(d).

 

Fig. 5. Results were obtained for baseband system operation, without the 20 GHz carrier. We note the shape of the driving 2.5 Gb/s data source signal (a); the output waveform from the optical transmitter (Measuring Point, MP1) (b); the corresponding DFB output (MP2) (c), and the resulting electrical waveform obtained after 40 Ghz photodetection and filtering with a 1.8 GHz Bessel LPF (MP3) (d)

Download Full Size | PDF

 

Fig. 6. Showing signal waveforms at various points in the system in response to a 20 GHz halfwave rectified ASK modulated signal input. The driving data waveform is identical to that of Fig. 5. We present details of the optical transmitter output waveform with different time resolutions (MP1) (a,b). The DFB output (MP2) and the output obtained by photodetection of this signal with a 40 GHz photodiode and filtering of the electrical signal obtained with a 1.8 GHz Bessel lowpass filter (MP3) are also shown as (c) and (d) respectively.

Download Full Size | PDF

Figure 6 presents the results obtained with ASK transmission using a 20 GHz carrier. The driving data source waveform is similar to that presented in Fig. 5; this information is not presented twice. We instead present a sample of the waveform from the optical transmitter (MP1), shown in Fig. 6(a). Figure 6(b) presents a more detailed view of this waveform, corresponding to a data transmission sample. The DFB output in response to this driving input is presented as Fig. 6(c); and the results obtained from photodetection and application of the lowpass filter (LPF) as previously described is shown in Fig. 6(d).

From Fig. 6, we observed that the output of the optical transmitter is characteristic of an optical halfwave rectified signal, such as would be obtained at the output of an all optical wireless signal detector [2], shown in Fig. 1. The DFB output waveform exhibits the overshoot and relaxation oscillations associated with the switching of this device. We have observed improved rise and fall times of the DFB laser output: in response to an input signal having rise time of 22.2 ps, the minimum rise time of the DFB output is 14.6 ps. We additionally noted the removal of DFB output overshoot and high-frequency oscillations at MP3. Logic inversion was observed between the electrical input signal waveform and the DFB output. It is straightforward that application of appropriate bias to the EAM at the system input corrects for this effect and preserves signal logic during system transmission.

 

Fig. 7. BER sensitivity for fixed P_TL = -7.5 dBm and fixed DFB bias current of 21.10 mA. The baseline performance (black line) is very close to the best performance with the envelope detection of the 20 GHz carrier, 2.5 Gb/s ASK passband signal (red). (Color online)

Download Full Size | PDF

4.3 BER sensitivity to received power

We evaluated bit-error rate (BER) sensitivity to receiver power with baseband and 20 GHz halfwave rectified input signal for fixed injected power level to the DFB and fixed DFB bias. Based on previous observations, the wavelength of the tunable source was set to λ_HR1 as this operating point corresponds to the maximum response of the DFB to the injected signal. The electrical data pattern consisted of PRBS of lengths (2⁷ − 1) and (2¹⁵ − 1) bits.

It may be seen from Fig. 7 that the optical transmitter signal yields a receiver BER sensitivity of 10^-9 at approximately -37.7 dBm average received optical power. We consider this the baseline scenario for evaluation of the signal quality effects introduced by the optically-injected laser system. Results for BER sensitivity analysis with the DFB are presented in Table 1 for baseband (carrier off) and passband (ASK modulation of 20 GHz carrier) system operation. With a PRBS of length (2¹⁵-1) bits, we observed that the system sensitivity (at 10^-9 BER) increased by 0.6 dB when the 20GHz carrier wave was activated in the ASK injected optical signal. With PRBS of length (2⁷-1) bits, we observed that the system performance with envelope detection of the input optical signal was associated with a BER sensitivity degradation of 0.1 dB, relative to the sensitivity of the optical signal into the DFB. The worst-case receiver power penalty associated with inserting the optically-injected system is less than 1 dB relative to the baseline situation at this BER. This value may be considered to lie within tolerable experimental error limits.

These results indicate that negligible performance penalty is associated with insertion of the optically injected DFB system at the bias point and optical injected power selected. For DFB bias current close to 21.1 mA, the results of Fig. 4 indicate that optical power levels above the threshold P_TL = -4 dBm cause deviation of DFB output wavelength away from λ_C. We noted that the selection of average optical injection power level P_TL = -7.55 dBm into the DFB provided 3.5 dB operating margin before the onset of this wavelength shift.

Table 1. Receiver sensitivity at BER = 10^-9 for various PRBS with 20 GHz carrier ON or OFF^a

View Table

4.4 BER sensitivity to optical power level into DFB, P_TL

We then assessed the relationship between receiver sensitivity and injected optical power level P_TL into the DFB when the system performed all-optical envelope detection of a 2.5 Gbit/s waveform ASK-modulated onto a 20 GHz carrier. For this assessment, the DFB bias current was fixed at I_DFB = 21.10 mA: the corresponding results are presented in Fig. 8. From Fig. 8, the following trends can be observed:

We note that a sensitivity of less than -37 dBm for a target BER of 10^-10 may be achieved for:

4.5 Assessment of wavelength dependency of XGM effect produced by optical injection

We characterized the wavelength dependency of the injection locking effect used in this system. The first assessment was performed by maintaining a constant input power into the DFB and varying the injected signal wavelength. The response of the DFB was evaluated in terms of the output power level and the output spectrum. We observed very similar response of the DFB device when the wavelength of the injected light corresponded to any of the spectral peaks in the range indicated in Fig. 2. This suggested that the envelope detection effect exploited throughout this paper is applicable across a wider range of input wavelengths. The DFB response to injected signal was negligible (i.e. no gain compression was observed) when the wavelength of the injected signal coincided with any of the minima (valleys) of the spectral curve as shown in Fig. 2.

 

Fig. 8. Sensitivity analysis of DFB system (for I_DFB = 21.10 mA). These results were obtained when the injected signal was at a wavelength λ_TL = λ_HR1

Download Full Size | PDF

 

Fig. 9. Showing the eye diagrams obtained at system output with increased data rates of 4 Gbit/s (a) and 5.5 Gbit/s (b). A 4.5 GHz Bessel LPF was implemented after photodetector.

Download Full Size | PDF

We then assessed the DFB response to an optical signal at an intermediate point between one of the peaks and valleys indicated in Fig. 3(a), to evaluate the power penalty incurred by operating the optically-injected system with misalignment between the injected signal and the DFB spectrum. For this analysis, we considered operation near the sixth harmonic wavelength greater than λ_C, denoted λ_HR6.

We observed the sixth peak at a wavelength of λ_HR6 = 1559.116 nm, the corresponding power output was -56.08 dBm. We selected λ_TL = λ_HR6 and P_TL = -7.55 dBm and noted the minimum receiver sensitivity at target BER = 10^-9. We varied λ_TL around λ_HR6 (with this fixed value of P_TL injected into the DFB) and observed that the XGM-dependent attenuation of the DFB output was decreased (i.e. DFB output power increased) when λ_TL deviated away from λ_HR6.

In order to determine the power penalty paid at the optical injection source due to this wavelength detuning, the DFB output was assessed when λ_TL - λ_HR6 = ε ≠ 0. We selected an injection wavelength of λ_TL = 1559.240 nm (therefore ε = 124 pm) and varied P_TL until the receiver sensitivity returned to the value noted in the previous paragraph. Under these conditions, it was found that P_TL had to be increased by 1.7 dB to achieve the same receiver sensitivity. Analysis of the DFB spectrum shows that the difference in output power obtained at this pair of wavelengths is 1.5 dB. The power penalty of misaligned wavelength operation is therefore greater than output power difference by 0.2 dB, which is less than 14% of the difference in DFB output power.

4.6 Assessment of transmission at increased bit-rate

We implemented a 4.5 GHz Bessel LPF, increased the data rate through the system and evaluated eye diagrams obtained at MP3. Eye diagrams are presented in Fig. 9(a) for operation at 4 Gbit/s, and in Fig. 9(b) for 5.5 Gbit/s. Full system characterization was not evaluated for this test, although these results indicate potential to operate at higher bitrates.

5. Conclusion

We have presented experimental results demonstrating that gain suppression due to external optical injection of a DFB laser can be exploited to perform envelope detection of wireless signal in a radio-over-fiber communications scheme. A characterization of the DFB output signal as a function of optical injected power level and wavelength values has been presented, leading to the determination and characterization of a suitable operation point for this system.

We assessed the performance of the optically-injected device under static conditions in which the wavelength and power of the injected optical signal is slowly varied and present observations made of the corresponding DFB output power and spectrum. Advanced system characterization is also presented for system operation at 2.5 Gb/s under baseband and passband conditions. Data transmission was experimentally evaluated with (2⁷-1) and (2¹⁵-1) bit-length PRBS: passband conditions were evaluated using ASK modulated 20 GHz carrier.

With the all-optical envelope detection scheme evaluated, optical signals yielding receiver sensitivity better than -37 dBm were achieved at a target 10^-10 bit error rate (BER) with an average optical injected power less than -7.5 dBm. The BER sensitivity of the optical signal output from the DFB very closely followed that of the data source. These encouraging results indicate that the proposed all-optical envelope detection for wireless signals may provide benefits for accelerated convergence of radio-over-fiber backbone as well as fixed broadband optical communications systems.

We note that this novel method for all-optical envelope detection, using an opticallyinjected laser device, has been successfully demonstrated to operate under conditions similar to those suitable for future radio-over-fiber applications. We observe that the simplicity of this scheme, as well as the wide range of operating conditions available, give this scheme significant potential for implementation in current and future systems.